FreeCalypso status update

[Mychaela Falconia]

Hello FreeCalypso community,

I have done some more polishing on our Magnetite hybrid firmware, and
I am pleased to report that this hybrid fw for AT-command-controlled
modem functionality now works no worse than our previous stable
firmwares that relied on binary-only libraries for essential GSM
functionality.  The new hybrid fw is built almost entirely from
source, containing no blobs other than Nucleus and some small parts of
GPF (we already have source replacements for those, but they have yet
to be integrated into the Magnetite->Selenite line of development),
and has been achieved by using new versions of G23M PS and ACI
firmware components from TI's TCS3/LoCosto program.  The new PS+ACI
combo initially had a few minor bugs and missing features compared to
our previous stable firmwares, but now that the visible bugs have been
fixed and the visible missing features have been added, the new fw is
in a state where I say it is ready to be beta-tested in real-life
usage.

The point needs to be repeated: we now have fully functional, fully
operational GSM+GPRS mobile station firmware with voice, SMS, CSD and
GPRS functionality in which all GSM-affecting and radio-affecting code
is fully built from source, and it works reliably to the extent of my
ability to test it in my lab on an FCDEV3B board connecting to the
live commercial GSM network of T-Mobile USA.

There is not much work remaining to be done on the sw/fw side: I will
need to dig up the OSL source reconstruction which I did back in 2014
(used in Citrine fw), polish it a little better and integrate it into
Magnetite so we can switch the GPF component from prebuilt binary libs
to recompilation from source, then swap out Nucleus (replace it with
the source version currently used in Citrine), and then finally
transition to compiling with gcc instead of TI's proprietary TMS470
compiler.

With the sw/fw essentially done, what we need now in order to put this
full-source-enabled GSM+GPRS modem solution into the hands of end
users is hardware.  Calypso chips are still readily available on the
Chinese surplus market (at least tens if not hundreds of thousands of
pieces), but unless you are going to eat them, chips by themselves
aren't of any use to an end user: we also need board-level hardware
built from those chips.

On the hardware front, the 3 FCDEV3B V1 boards which I had available
for sale have just been sold, thus the boards are currently sold out:
I now have just two boards left from the good batch, and I need them
for my own development and testing, hence they are not for sale.

So how can we put our FreeCalypso GSM solution into the hands of Joe
Free Software End User?  I see several possible ways, although each of
them will require significant money and time (the time can be shortened
given a lot of money and vice-versa):

Approach 1: build a FreeCalypso dumbphone handset, a phone in which
the Calypso would act as the main and sole processor as opposed to a
slave modem.  This is my personally preferred solution, but it has the
disadvantage of being absolutely unattractive to those who have no
interest in dumbphones.

Approach 2: repackage our FC modem into the form factor of an SMT
module similar to the mainstream proprietary ones which the whole
world is using, and then convince someone to build a libre smartphone
or some other kind of end user product with that module.  Practically
speaking, this approach would only work if someone else pays for it:
I am not willing to spend tens of thousands of dollars of my own money
to produce the repackaged modem module only to subsequently have zero
success in convincing anyone to use it.

Approach 3: repackage our FC modem into a USB-powered dongle, so it
can be connected directly to USB without needing an intermediate
adapter board like we currently use with FCDEV3B or a separate non-USB
power supply.  This option would be a direct end user product, i.e.,
it could be marketed directly to end users without having to find some
third party willing to build a smartphone or some other complex product
with our modem.

However, before we can do any of the above, the first required step is
to make more FCDEV3B boards from the new design version I am calling
FCDEV3B V2.  The new board revision has the flash reset signal wired
differently as discussed previously on this list, and the change is
expected to fix the sleep mode bug (allow sleep modes to be used with
firmware running from flash) and also minimize power consumption
during sleep.  (Our flash chip should NOT be held in reset to get the
lowest standby current draw.)

The current status of FCDEV3B V2 is that a preliminary version of the
PCB design is already done, we just need to add one more test point
for safety, and then we are going to need somewhere around 4-5 kUSD to
get the boards physically made: PCB fab run, a new parts order, a new
paste stencil and an assembly run.

The money from the sale of the last 3 FCDEV3B V1 boards will go toward
paying John in Colorado (my Altium PCB contractor) for the work of
producing the PCB design for FCDEV3B V2, but we still need someone
else to pay for the next step of PCB fabrication, expected to be
somewhere in the $2600 to $3000 range depending on the number of
panels we order.  Thus that is the next step - we need to get someone
to donate that much money in order to move forward.

Hasta la Victoria, Siempre,
Mychaela aka The Mother